From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Thu Jun 05 2014 - 17:27:37 CDT
It is my understanding that these guys at the Benoît Roux lab (including
Sergei Noskov who worked for him at the time) must have spent man-years
fine-tuning what amounts to a huge cancellation of errors for these ions
in toppar_water_ions. While some might say the parameters will never be as
good as other parts of the force field because of lack of charge transfer
in the Potential Energy Function, they're as good as they get, and should
be pretty acceptable for a lot of purposes. Interactions with anionic
groups might still pose problems as long as there are no specific NBFIX
terms for them, especially for the smaller ions such as Mg. That may seem
counterintuitive, but *if* I understand correctly (that's a big "if"), the
small radius on the smaller ions allows the positive charge to get very
close to the negative charges on the anion, and yield an electrostatic
interaction that is an overestimation of the actual interaction. For large
cations, the problem seems to be opposite; there's often a large dipolar
(aka. coordinate covalent) bonding character, which is not captured by the
force field, and which leads to underestimation as well as lack of
directionality.
As to answer your question (as good as I can; I'm no ions expert):
- Na+ has been studied to death and has NBFIXes, so should be pretty good.
- K+ has only a single positive charge and *may* not be as much in need of
NBFIXes, though I've been told to preferentially neutralize with Na+ if I
have the choice, just to play it safe.
- Ca++ and Mg++ become a bit more problematic because there's so much more
charge transfer. I'd especially expect Mg++ to stick far too strongly to
carboxylates by lack of NBFIXes; Ca++ *might* do better. Don't take my
word for it, though; these trends are not always as logical as one would
expect and I don't have first-hand experience.
- because of all this, you should be very skeptical about the occupancy
and residence time in binding sites that contain solvent-exposed negative
groups; conversely, I would give your proposed study a reasonable chance
of success if you focus on the non-anionic binding sites.
- Anything that form chelate complexes (such as transition metals) is
highly suspect.
- I'd say, keep us posted if you find more concrete information in the
literature.
On 06/04/2014 05:00 PM, Aron Broom wrote:
> Thanks for the link Kenno, I had actually been wondering about doing
> something to compare binding sites for Na+ with Mg++, but seeing the
> struggle people have had puts the matter into a better light.
>
> I wonder though if ions like Mg++ and Ca++, might be less horrific than
> anything involving Fe, Cu, Zn, or the Au Daniel is trying for? The thread
> you linked did suggest some improvement with polarizability, so I wonder
> if ions that are dominated by simpler S orbitals and can be treated more
> like a sphere would do better than the transition metals? Of course that
> wouldn't help in the case of this gold ion, which seems like it might be
> particularly troubling.
>
>
>
>
> On Wed, Jun 4, 2014 at 11:32 AM, Kenno Vanommeslaeghe
> <kvanomme_at_rx.umaryland.edu <mailto:kvanomme_at_rx.umaryland.edu>> wrote:
>
> I know you said "SUPER rough", but even so, I'm strongly opposed to
> the idea of adjusting a radius without the well depth (or adjusting
> both while targeting only one value such as an estimated size). Thin
> ice, there; then it *might* be more accurate to take a potential for a
> neutral gold atom and just add the +1 charge, even though that's also
> very, very wrong (not only do the L-J parameters depend strongly on
> the formal charge, atoms at the surface of bulk metal have strongly
> different vdW behavior than isolated ones).
>
> Also, if the physical system one wants to simulate has Au(3+), doing a
> simulation with Au(1+) won't be of much help; they'll behave totally
> different.
>
> There simply does not exist an easy solution to the problem at hand.
> In the words of one of the top experts in CHARMM ion parametrization:
> http://www.charmm.org/__ubbthreads/ubbthreads.php?ubb=__showflat&Number=31423#__Post31423
> <http://www.charmm.org/ubbthreads/ubbthreads.php?ubb=showflat&Number=31423#Post31423>
> What came before in that thread is also somewhat of a sobering read...
>
> Axel's advice is very insightful and thorough - I couldn't possibly
> improve upon it.
>
>
>
>
> On 06/03/2014 05:21 PM, Aron Broom wrote:
>
> I think if you want to do something SUPER rough as you are
> suggesting with
> an Au1+, you could use existing CHARMM or AMBER forcefields and
> use Na+ or
> K+ and adjust the VDW coefficient to match some estimate of the
> size of Au1+.
>
> But keep in mind that in the simple case of classical MD, those
> atoms are
> just point charges with some VDW radius, which does a poor job of
> representing metal ions, particularly ones that have more complex
> orbitals
> than a sodium. So really, you'll just be seeing where a big
> sodium would
> sit on your protein, which will potentially be a lot different
> than the
> actual case. In fact, if you are thinking of doing that, you'd
> almost be
> better advised to just do some kind of Poisson-Boltzman
> calculation of the
> electrostatic surface potential of your protein and see where the
> negative
> potentials are distributed.
>
> As Axel suggests, you'd need some more complex treatment to get
> anything
> accurate. Metals are hard.
>
>
>
>
>
> On Tue, Jun 3, 2014 at 4:49 PM, Axel Kohlmeyer <akohlmey_at_gmail.com
> <mailto:akohlmey_at_gmail.com>
> <mailto:akohlmey_at_gmail.com <mailto:akohlmey_at_gmail.com>>> wrote:
>
>
>
> On Tue, Jun 3, 2014 at 10:17 PM, Daniel Torrente
> <xlb608_at_my.utsa.edu <mailto:xlb608_at_my.utsa.edu>
> <mailto:xlb608_at_my.utsa.edu <mailto:xlb608_at_my.utsa.edu>>> wrote:
>
> Thanks for your quick response
>
> I will take your advice and keep looking for published
> literature
> on this topic. In the mean time, if I tried to use Au+ or Au
> instead of Au3+, is there any ff available that could
> help to
> carried out this simulation?. Basically, I do not want to
> fix the
> Au atom. I want them to move freely during the simulation
> and see
> the aggregation pattern over the surface of the protein.
>
>
> again, i think a "just give me the force field" kind of
> approach is
> going to be successful here. there is much more literature and
> fundamental research ahead of you before i would even begin
> any kind
> of simulation. you first have to understand the chemistry of
> what you
> are looking at well enough.
>
> that will most likely require some level of ab initio or
> semi-empirical calculation. using a subset of your total
> system to
> train and benchmark any force field parameter set.
> i would suspect that you still need some form of many-body
> potential,
> eg. MEAM to represent Au atoms and study their clustering, which
> means you would have to use an MD code that can handle
> multiple force
> field types at the same time or supports (mechanical or better)
> coupling of codes.
>
> axel.
>
>
> Thanks
>
> Daniel Torrente
>
>
>
>
> On Tue, Jun 3, 2014 at 2:38 PM, Axel Kohlmeyer
> <akohlmey_at_gmail.com <mailto:akohlmey_at_gmail.com>
> <mailto:akohlmey_at_gmail.com <mailto:akohlmey_at_gmail.com>>>
> wrote:
>
>
> hi daniel,
>
>
> On Tue, Jun 3, 2014 at 8:31 PM, Daniel Torrente
> <xlb608_at_my.utsa.edu <mailto:xlb608_at_my.utsa.edu>
> <mailto:xlb608_at_my.utsa.edu <mailto:xlb608_at_my.utsa.edu>>> wrote:
>
> Hi guys,
>
> Is there any available force field that can
> simulated the
> interaction between Au+3 and a protein? I was
> looking for
> some information in articles and the mailing
> list related
> to this topics, but all I could find was the GoiP and
> GoiP-charmm (immobile surface Au). Also found the
> charmm-metal ff but it seems that only works with the
> metal Au and not with the ion Au+3 (there is no
> information in the ff about Au+3).
>
> Any suggestion on how to approach this type of
> interaction? or or could I do this with any of
> the ff that
> I mentioned before?
>
>
> multiply charged ions are very problematic for pair-wise
> additive force fields as their interactions usually
> include
> polarization of the immediate environment with charge
> redistribution and directional interactions. most
> likely you
> will not have a "naked" au3+ cation, but some kind of
> complex
> with nulcleophilic molecules and/or anions. depending
> of what
> you want to study, you may need to resolve to doing QM/MM
> calculations, or parameterize an au3+ complex that
> you keep
> rigid or otherwise maintain its charge distribution.
>
> interactions with gold surfaces are a different
> system, since
> those are usually dominated by a mostly covalent bond
> (via
> sulphur or oxygen) and the effects due to
> polarization of the
> metal are smaller than other errors of the model to be
> justifiably ignored. often people use rather crude
> models for
> it, since they don't care as much about the
> interaction with
> the gold surface than of the objects attached to the
> surface
> with each other and the items around it.
>
> i suggest you have another look at the published
> literature
> and think carefully what it is that you really want
> to learn
> from your simulations and come back, if you still
> have questions.
>
> also, please don't just take a single opinion on the
> subject
> as your guideline. for anything this problematic, you
> have to
> look at multiple contrasting opinions and form your own
> opinion as there is no single simple answer that
> answers all
> problems.
>
> axel.
>
>
>
>
> Thanks in advance
>
> Daniel Torrente
>
>
>
>
> --
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com
> <mailto:akohlmey_at_gmail.com>
> <mailto:akohlmey_at_gmail.com
> <mailto:akohlmey_at_gmail.com>> http://goo.gl/1wk0
> College of Science & Technology, Temple University,
> Philadelphia PA, USA
> International Centre for Theoretical Physics,
> Trieste. Italy.
>
>
>
>
>
> --
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com
> <mailto:akohlmey_at_gmail.com> <mailto:akohlmey_at_gmail.com
> <mailto:akohlmey_at_gmail.com>>
> http://goo.gl/1wk0
> College of Science & Technology, Temple University,
> Philadelphia PA, USA
> International Centre for Theoretical Physics, Trieste. Italy.
>
>
>
>
> --
> Aron Broom M.Sc
> PhD Student
> Department of Chemistry
> University of Waterloo
>
>
>
>
>
> --
> Aron Broom M.Sc
> PhD Student
> Department of Chemistry
> University of Waterloo
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:27 CST